B2B SAAS | INTERNSHIP
AI Assistant Context Orchestration
Making the invisible visible for enterprise AI 👓
ROLE
Product Design Intern (AI Team)
TEAM
Me (main owner)
2 Designer Managers
1 User Researcher
3+ Developers
1 PM
TIMELINE
Summer 2025
SKILLS
AI Agent Design
Product Design
Visual Design
Design Systems
OVERVIEW
What would Splunk’s AI look like if it truly understood its users?
Splunk (a Cisco company) is an enterprise platform that turns complex machine data into actionable insights through dashboards, alerts, and visualizations.
During my internship at Splunk, I focused on improving its AI Assistant by making the system’s behind-the-scenes process more transparent.
PROBLEM
Users lack visibility into the context used to generate AI responses.
Splunk AI Global Assistant is a high-priority project and already MVP-ready. However, previous research shows that a lack of context often leads to user confusion and difficulty, especially for AI loading.
😖
😨
🤔
Users had no insight into what data, pages, or sources the AI referenced while generating responses. Additionally, when responses took up to two minutes, the lack of transparency led to user uncertainty and frustration.
SOLUTION
Turning loading into a story users can follow, through two layers of visibility.
Now, when users ask questions to the AI Assistant, they see more than just a spinning icon and the word “Processing.” They also receive:
01
Traces of tool invocation
a step-by-step view of the AI’s actions during response generation.
02
In-depth context cards
Detailed breakdowns of the inputs and environment data that shaped the AI’s reasoning.
Due to NDA, the project detail cannot be shared. Contact me at yl3262@cornell.edu to learn more!
IMPACT
Rippling effect across the company…
TESTING
Delivered "User Testing Insights One-Pager", where design received user satisfaction score 4.4 out of 5 from 7 real customers.
PATTERN
Defined engineering-ready design patterns for transparent AI orchestration loading state across multiple products.
ROADMAP
Laid the groundwork for future AI agent design while securing stakeholder alignment and integrating the work into the FA26 roadmap.
REFLECTIONS
First AI focused project, my takeaways:
AI UX is narrative, not just functional🗣️
Users want to see the reasoning, not just the results. Future AI must present processes as stories to build trust!
Scalability requires a systems mindset 🔧
Scaling design means anticipating dependencies, edge cases, and evolution. UX must adapt fluidly across platforms and contexts.
Transparency matters for both AI and people 👓
Visibility into decisions and open communication are key. Transparency fosters trust and accountability.
Photo Gallery @ Splunk
Last day photo at the Boulder office!
Regular meeting with my intern manager Eli and engineering stakeholders
